Document reRvjmydRk9QzRyjjErdXDKmV

Available online at www.sciencedirect.com SCIE NC E DIRECT Regulatory Toxicology and Pharmacology 39 (2004) 363-380 Regulatory Toxicology and Pharmacology www.elsevier.com/locate/yrtph Characterization of risk for general population exposure to perfluorooctanoate John L. Butenhoff,a * David W. Gaylor,b John A. Moore,c Geary W. Olsen,a Joseph Rodricks,d Jeffrey H. Mandel,e and Larry R. Zobela a Medical Department, 3M Company, Building 220-2E-02, St. Paul, M N 55144, USA b David W. Gaylor Associates, Little Rock, AR, USA c Hollyhouse, Inc., Wicomico Church, VA, USA d Environ Health Sciences Institute, Arlington, VA, USA e Exponent, Chicago, IL, USA Received 10 November 2003 Available online 16 April 2004 Abstract Perfluorooctanoate (PFOA), an environmentally and metabolically stable perfluorinated carboxylic acid, has been detected in the serum of children, adults and the elderly from the United States with the upper bound of the 95th percentile estimate in the range of 0.011-0.014 pg/mL (ppm). In this risk characterization, margins of exposure (MOE), which can provide a realistic perspective on potential for human risk, were determined by comparison of general population serum PFOA concentrations with serum con centrations from toxicological studies that are associated with the lower 95% confidence limit of a modeled 10 percent response or incidence level (LBMIC10) using USEPA BMDS software. The LBMIC10 was estimated using surrogate data from other studies or pharmacokinetic relationships if serum PFOA data were not available. Modeled dose-responses (with resulting LBMIC10 values) included post-natal effects in rats (29 pg/mL), liver-weight increase (23 pg/mL), and body-weight change (60 pg/mL) in rats and monkeys, and incidence of Leydig cell adenoma (125 pg/mL) in rats. MOE values based on the upper bound 95th percentile population serum PFOA concentration were large, ranging from 1600 (liver-weight increase) to 8900 (Leydig cell adenoma). These MOE values represent substantial protection of children, adults, and the elderly. 2004 Elsevier Inc. All rights reserved. Keywords: Perfluorooctanoate; Perfluorooctanoic acid; C8; PFOA; APFO; Risk characterization; Biomonitoring; Benchmark dose; Benchmark internal concentration; Toxicokinetics 1. Introduction Perfluorooctanoic acid is a fully fluorinated carbox ylic acid that, due to the strength of the carbon-fluorine bond, is exceptionally stable to metabolic and environ mental degradation. The presence of fluorine in the carbon chain imparts a high electron-withdrawing ca pacity, rendering the carboxyl function highly acidic relative to other organic acids. Salts of perfluorooctanoic acid have been used as surfactants and processing aids in the production of fluoropolymers, and these salts Corresponding author. Fax: 1-651-733-1773. E-mail address: jlbutenhoff@mmm.com (J.L. Butenhoff). 0273-2300/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2004.03.003 are considered critical to the production of certain flu oropolymers and fluoroelastomers. Perfluorooctanoate (PFOA), the dissociated anion of the acid, has been found in the serum of children, adults, and the elderly as part of a broad biomonitoring study of the United States (US) population (Olsen et al., 2003c, 2004a,b). Serum concentrations of PFOA follow a log normal distribution with geometric mean concentrations of 0.004-0.005 pg/mL. PFOA was quantifiable in over 90% of the serum samples. Upper 95% confidence limits of the 95th percentile estimated serum PFOA concentra tions ranged from 0.011 to 0.014 pg/mL (Table 1), and the highest measured individual values in these general populations were 0.056, 0.052, and 0.017 pg/mL for children (ages 2-12), adults (ages 20-69), and the elderly 364 J.L. Butenhoff' et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 Table 1 Serum PFOA concentrations measured in three non-occupationally exposed populations from the United States General population monitoring study Location Sample sizee Upper bound o f the 95th percentile estimate o f the population (pg/mL) Childrena Adultb Elderly0 23 states 6 citiesd Seattle (WA) 645 598 238 0.011 0.014 0.011 aOlsen et al. (2004a). bOlsen et al. (2003c). cOlsen et al. (2004b). dPortland (OR), Los Angeles (CA), Minneapolis-St. Paul (M N), Charlotte (NC), Hagerstown (M D), and Boston (MA). eSample size in each study was equally represented by sex. Geometric mean serum PFOA concentrations were similar by sex. (ages 65-96), respectively. Although possible sources and pathways of exposure have been suggested (environ mental releases and some consumer products), the source(s) and pathway(s) responsible for human exposure are not known. The United States Environmental Protection Agency (USEPA) and current and former manufacturers and users of PFOA and its salts have joined in a process to better understand sources and pathways of PFOA expo sure (USEPA, 2003a). Based on interim results from a study of nine retired workers, PFOA appears to be poorly excreted in humans, as the current estimate of serum elimination half-life from an on-going study of retired workers is 4.4 3.5 years (Burris et al., 2002). The finding of PFOA in most human sera samples tested and the extended retention time in the body have prompted consideration of the potential health risk of low-level continuous exposure to PFOA. The potential toxicity of PFOA has been studied ex tensively, and recent reviews of PFOA toxicity are available (Kennedy et al., 2004; USEPA, 2002). While many of these studies have been published or are in various stages of preparation for publication, the USEPA administrative record 226 (AR-226) contains copies of many original studies not yet available in the peer-reviewed literature, as well as source data for sev eral published studies. Most studies were conducted with the ammonium salt of PFOA, which readily dis sociates in aqueous media at physiological pH. The designation, APFO, will be used when referring specifi cally to this salt. Numerous acute, shorter-term, and longer-term toxicity studies have been conducted in multiple species by different routes of exposure. In ad dition, two chronic dietary studies in rats have been reported. The toxicological database includes develop mental toxicity, reproductive toxicity, immunotoxicity, genotoxicity, carcinogenicity, pharmacokinetics, and various mode-of-action studies. In addition to toxico logical studies, medical surveillance and epidemiological studies of PFOA-exposed workers at the 3M Company have been ongoing since the late 1970s (3M Company, 2003a,b; Alexander, 2001; Gilliland, 1992; Gilliland and Mandel, 1996; NIOSH, 2001; Olsen et al., 1998, 2000, 2003a; Ubel et al., 1980). PFOA should not be confused with perfluorooctanesulfonate (PFOS, C8 F17SO3), another eight-carbon perfluorinated acid that has also been found to be eliminated slowly in humans (Burris et al., 2002), is widely distributed in humans (Olsen et al., 2003c, 2004a,b), and has been found in wildlife samples from many areas of the world (3M Company, 2003c). There are key differences in the production, use, environmental distribution, potential toxicity, and pharmacokinetics between PFOA and PFOS; therefore, these two perfluorinated acids should be evaluated separately. There have been two recently documented pre liminary risk assessments focused on PFOA. In the first, the State of West Virginia Department of Environ mental Protection released a report on the establishment of preliminary risk screening levels based on RfD and RfC values derived by a team of experts for PFOA in drinking water, soil, and air in proximity to a manu facturing facility that uses APFO in fluoropolymer production (West Virginia Department of Environ mental Protection, 2002). All water samples collected in the vicinity of this facility were below the risk screening level of 150 qg/L derived for drinking water in this process. Water samples from 50 private well and cisterns used for drinking water and the nine public water sup plies were below 3 qg/L. No measured air or soil con centrations were available. Another preliminary risk assessment was released by USEPA (USEPA, 2003b), which presented a range of margin-of-exposure (MOE) values based on compari sons of human serum concentrations of PFOA (Olsen et al., 2003c, 2004a,b) and the serum concentrations in samples taken from rats involved in a two-generation reproduction study. However, the USEPA cautioned that these MOE values should not be considered to represent the range of possible MOE values for general populations because of uncertainties resulting from the lack of appropriate toxicokinetic data in weanling rats and their relationship to human serum levels of PFOA. The USEPA continues to develop a more comprehen sive risk characterization that is intended to better define potential human MOE values (Butenhoff et al., 2004; USEPA, 2004). J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 365 This document presents a characterization of poten tial health risk for the general population using mea sured serum PFOA concentrations in the biomonitoring studies conducted by Olsen et al. (2003c, 2004a,b). Dose-response data from toxicological studies have been used to estimate concentrations of PFOA in serum associated with a 10% benchmark response (BMR) for several key endpoints. Further, the lower 95% confi dence limits of these benchmark internal concentrations (LBMIC10, Gaylor et al., 2003) have been used as a basis for comparison with general population serum PFOA concentrations. This method takes advantage of the facts that: (1) PFOA is chemically stable and not readily subject to environmental and metabolic degradation (Goecke et al., 1992; Kuslikis et al., 1992; Vanden Heuvel et al., 1991). (2) Extensive toxicological (including non-human pri mates) and worker-health studies are available (Kennedy et al., 2004; USEPA, 2002) that allow for examination of most toxicological endpoints (e.g., developmental toxicity, reproductive toxicity, immunotoxicity, genotoxicity, carcinogenicity, phar macokinetics, and mode-of-action). (3) Serum PFOA measurements have either been made in connection with studies or can be estimated based on new information on the toxicokinetics of PFOA in the rat (Han, 2003; Kemper, 2003; Mylchreest, 2003) and monkey (Kerstner-Wood et al., 2003; Noker, 2003) to facilitate cross-species extrapolation. (4) Population exposure to PFOA has been well charac terized through serum PFOA concentration mea surements in biomonitoring studies that include children (Olsen et al., 2004a), adults (Olsen et al., 2003c), and the elderly (Olsen et al., 2004b). The use of these factors can be instrumental in re ducing uncertainty in the risk characterization of PFOA. 2. Methods 2.1. Selection o f studies and endpoints A review of the toxicological database for PFOA was conducted in order to select studies that covered a va riety of endpoints, were sufficiently robust, and provided good dose-response data. The endpoints and associated studies chosen are presented in Table 2. Sensitive indi cators of response that were chosen for the determina tion or estimation of benchmark internal concentration values (LBMIC10, as described in Section 2.3) were post natal developmental effects (rats), liver-weight increase Table 2 Endpoints and source studies used in evaluating dose-response Endpoint Source study Source data table Post-natal development in ratsa Liver-to-brain-weight ratio in ratsb Body-weight change in ratsc Liver-to-brain-weight ratio in ratsd Body-weight change Liver-to-brain-weight ratio in monkeys' Body-weight change in monkeysf Leydig cell tumors in rats8 Two-generation reproduction study (Butenhoff et al., 2004) Two-generation reproduction study (Butenhoff et al., 2004) Two-generation reproduction study (Butenhoff et al., 2004) 13-week dietary study (Palazzolo, 1993) 13-week dietary study (Palazzolo, 1993) 6-month oral toxicity study (Butenhoff et al., 2002b) 6-month oral toxicity study (Butenhoff et al., 2002b) Two-year cancer bioassay (Sibinski et al., 1983) Table 3 Table 3 Table 3 Table 4 Table 4 Table 5 Table 5 Table 6 aThe following endpoints were evaluated separately: (1) pre-weaning mortality (combined sexes); (2) pup body-weight at weaning (combined sexes); (3) post-weaning mortality in males and females (separately); (4) days to preputial separation in males; and (5) days to vaginal patency in females. bMale liver-weight-to-brain-weight ratio was selected because male rats respond to a greater extent than females to the liver-enlarging effects of PFOA. PFOA affects body weight; therefore, use o f liver-weight-to-brain-weight ratio normalizes for body-weight changes, since brain is not re sponsive to body-weight change from dietary restriction (Feron et al., 1973). F0 and F 1 data were evaluated separately. The two-generation re production study involved oral dosing o f male rats in both the F0 and F1 generations for more than 90 days, the typical term o f a subchronic study, and, therefore, has the advantage o f following a subchronic dosing response over two generations and group sizes o f approximately 30. cBody-weight change was evaluated as reduced body-weight gain compared to controls only in male rats, which were more sensitive than female rats to PFOA-induced reductions in weight gain. F0 and F1 data were evaluated separately. dLiver-weight-to-brain-weight ratio was used to minimize effects o f body-weight reduction and reduced feed consumption. The 13-week (90-day) subchronic dietary study in male rats (Palazzolo, 1993) is useful in that serum PFOA concentrations were made at all dose levels. e Since the male monkeys from this study varied in age and weight at the beginning o f the study, and dosing with APFO caused significant weight loss among the high-dose-group monkeys, only data from male monkeys dosed until terminal sacrifice were used, which excludes data from three high-dose-group monkeys for whom dosing was suspended. f For male cynomolgus monkeys, body-weight change was represented by the actual percentage change in individual body weight from pre-study baseline weight through weight at or near termination (scheduled or unscheduled) o f dosing. Because these were adult monkeys o f various ages and weights, and due to the fact that only two o f six monkeys were dosed continuously for six months at the high dose, percent change in body weight from baseline was considered more meaningful than comparison o f body-weight change or terminal body weight between treated and control groups. gHuman epidemiological studies have not shown statistically significant associations o f exposure to PFOA with increased cancer mortality risk (Alexander, 2001). Leydig cell adenoma incidence from the two-year cancer bioassay in rats was used. 366 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 (rats and monkeys), body-weight change (rats and monkeys), and incidence of Leydig cell adenoma (rats). Source data are presented in Tables 3-6. A comment regarding the choice of Leydig cell ade noma incidence from a two-year cancer bioassay in rats (Sibinski et al., 1983) is in order. Human epidemiological studies have not shown associations of exposure to PFOA with increased cancer mortality risk (Alexan der, 2001). The two-year cancer bioassay in rats by Sibinski et al. (1983) produced significant increases in Leydig cell adenoma (males) and mammary fibroade noma (females). The mammary fibroadenoma incidence Table 3 Dose-response data for post-natal developmental endpoints in a two-generation reproduction study with ammonium perfluorooctanoate (APFO) in rats (Butenhoff et al., 2004) Post-natal effect Oral gavage dose level (mg/kg/day) 0 1 3 10 30 Post-weaning mortality, males, % Post-weaning mortality, females, % Mean days to preputial separation Mean days to vaginal patency Pre-weaning mortality, combined sexes, % Mean weight at weaning, both sexes, g F0 male liver-to-brain-weight ratio F1 male liver-to-brain-weight ratio F0 male body-weight change, g F1 male body-weight change, g 5.0 (3/60)a 0.0 (0/60)a 48.5 34.9 2.6 (10/385)a 37.4 9.0 1.2 (30)b 9.3 1.4 (30)b 400 37 (30)b 512 55 (30)b aIncidence is given in parentheses. bSample size (n) is given in parentheses. Statistically significant compared to controls (p < 0.05). 5.0 (3/60) 3.3 (2/60) 49.5 35.5 3.0 (11/372) 36.7 10.7 1.5 (30)* 10.8 1.5 (2 9 )* 395 46 (30) 473 52 (2 9 )* 5.0 (3/60) 1.7 (1/60) 49.4 34.1 4.4 (17/388)* 39.7 12.3 1.2 (30)* 1 2 .2 1.8 (3 0 )* 362 48 (30)* 468 50 (3 0 )* 3.3 (2/60) 1.7 (1/60) 49.7 34.8 2.5 (10/400) 38.8 12.8 1.8 (30)* 12.9 1.6 (30 )* 335 53 (30)* 445 58 (30)* 12 (7/60) 10 (6/60)* 52.2* 36.6* 6.7 (26/388)* 35.7 12.5 1.4 (29)* 13.6 1.7 (2 9 )* 253 63 (30)* 388 37 (30)* Table 4 Dose-response data from a 13-week subchronic dietary study with ammonium perfluoroctanoate in rats (Palazzolo, 1993) Estimated dose (mg/kg/day) Serum [PFOA] (pg/ml) Liver-weight-to-brain-weight ratio Body-weight change (g) 0 (ad libitum) 0 (pair fed)b 0.06 0.64 1.94 6.5 <1 (10)a <1 (10) 7 1 (10) 41 13 (10) 70 16 (10) 138 34(10) a Sample size (n) is given in parentheses. bControl group pair-fed to high-dose group. Statistically significant when compared to ad libitum control. # Statistically significant when compared to pair-fed control. 9.03 1.20 (15) 7.64 0.774 (15) 8.19 1.56 (14) 9.41 1.33 ( 15) 10.8 1.96 (15) 1 2 .6 2 .8 8 (14)*# 339 34.1 (25) 296 17.8 (2 5 ) 343 33.1 (2 5 ) 348 37.8 (2 5 ) 327 42.2 (2 5 ) 290 57.4 (2 5 )* Table 5 Dose-response data from a six-month oral toxicity study o f ammonium perfluorooctanoate in male cynomolgus monkeys (Butenhoff et al., 2002b) Dose (mg/kg) [PFOA] seruma (pg/mL) Liver-weight-to- brain-weight ratio [PFOA] serumb (pg/mL) Body-weight change (%) 0 3d 10 30/20 0.16 0.15 (4)c 72 47 (4 ) 85 20 (4) 155 102 (2)e 0.934 0.074 (4) 1.34 0.23 (4)* 1.30 0.23 (4 )* 1.22 1.2 (2)e 0.21 0.14 (6) 72.1 44.4 (4) 81.3 25.2 (6) 284 212 (6) 17 6 (6) 13 8 (4 ) 15 5 (6) - 5 9 (6)* aMean serum values SDs from samples taken during weeks 20, 22, 24, and 26. bMean serum values SDs from single samples taken at termination o f dosing or mean o f multiple samples taken at termination o f dosing and/or within two weeks prior to termination o f dosing. cSample size (n) is given in parentheses. dIncludes data derived from monkey I05721M, which was sacrificed in week 20 on day 137. e Includes data derived from the two monkeys that were dosed until scheduled end o f dosing at 26 weeks. Does not include data from monkey I05724 that was humanely sacrificed in week 5 on day 29. Monkey I05724 had a body weight o f 3505 g, a liver weight o f 83 g, a liver weight % o f body weight o f 2.37, a liver weight % o f brain weight o f 1.48, and a serum concentration o f 822 pg/mL in week 4. Statistically significant when compared to controls (p < 0.05). J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 367 Table 6 Poly-3 procedure adjusted Leydig cell adenoma incidence rates for rats given ammonium perfluorooctanoate in diet for a lifetime Dietary dose level (ppm) Overall mean APFO intake (mg/kg/day)a Leydig cell adenomab 00 30 1.3 300 14 0/44 2/44 (4.6%) 7/48 (15%) Estimated L B M D ff LBM Djo= 100 ppm Estimated LBM IC10d LBM Djo= 4.8 mg/kg/day (10%) LBMIC1o= 125 i g PFOA/mL (10%) aEstimated from feed consumption and analysis o f diet for ammonium perfluorooctanoate. bThe values in the denominator represent the effective lifetime number o f rats at risk o f cancer based on the Poly-3 adjustment procedure (Bailer and Portier, 1988) and value shown is the corresponding age-adjusted tumor incidence rate. cThe lower 95% CI o f the benchmark dose based on the multistage model and a 10% incidence. dEstimated using the equation o f the trendline in Fig. 2 (y = 46.1 x x06347) to yield a LBMIC10 o f 125 pg PFOA/mL serum corresponding to 4.8 mg/kg/day. was reported to be within the range of reported spon taneous incidences for this tumor type. Recent evalua tion of historical control data from studies conducted at DuPont Haskell Laboratory and those offered on the animal supplier's (Charles River Laboratories) web site (http://www.criver.com/techdocs/tech_pdf/2001TOXDATA. pdf) indicate that this assertion is correct, and these benign mammary adenomas are not likely to be related to treatment (Butenhoff et al., 2002a). A mechanistic two-year dietary study in male rats reported by Biegel et al. (2001), in addition to finding an increase in Leydig cell adenoma (8/76 (11%) versus 0/80 (ad libitum con trol) and 2/78 (2.5%, pair-fed control)), found an in crease in pancreatic acinar cell adenoma/carcinoma (7/ 76 (9.2%) versus 0/80 (ad libitum control) and 1/79 (1.3%, pair-fed control)), and hepatocellular adenoma (10/79 (13%) versus 2/80 (2.5%, ad libitum control) and 1/79 (1.3%, pair-fed control)) at the single study dose of 300 ppm in diet. The single-treatment-level data avail able from the Biegel et al. (2001) study do not allow insight into the characteristics of the dose-response re lationships for tumor incidence. With respect to the liver tumors, the proposed mechanism of peroxisome-proliferator-activated-receptor-a (PPAR-a) activation sug gests that these tumors have questionable or no relevance to humans (Ashby et al., 1994; Bentley et al., 1993; Cattley et al., 1998). Therefore, because the Leydig cell adenoma incidence in the Sibinski et al. (1983) study (Table 6) was higher than or comparable to tumor in cidences in Biegel et al. (2001) and provided a means to model dose-response, it was used to develop LBMICio and MOE values for nonlinear cancer risk. The human relevance of Leydig cell tumors observed by Sibinski et al. (1983) as well as Leydig cell, and pancreatic acinar cell tumors observed by Biegel et al. (2001) remains uncertain (Ashby et al., 1994; Bentley et al., 1993; Biegel et al., 1995; Cattley et al., 1998; Clegg et al., 1997; Cook et al., 1992, 1999; Liu et al., 1996a,b). 2.2. Choice o f general population serum PFOA value for comparison to LBMICio valuesfrom toxicological studies Four studies that survey PFOA concentrations in serum samples from non-occupationally exposed popu lations in the US have been conducted. Separate studies of serum samples from children enrolled in a Group A Streptococcal clinical trial (N = 598), adult American Red Cross blood donors (N = 645), and dementia-free elderly (N = 238) from a prospective study of cognitive function have shown serum concentrations of PFOA averaging approximately 0.005 pg PFOA/mL with an upper bound of the 95th percentile estimate approxi mating 0.011-0.014 pg PFOA/mL, as shown in Table 1 (Olsen et al., 2003c, 2004a,b). In another study, Olsen et al. (2003b) examined a total of 30 human donor livers for the presence of PFOA. All donor livers were below the lower limit of quantitation for at least one of two analyses per sample except for one liver that had an average of 0.047 pg/g. The value from Table 1 that will be used to represent the upper bound of general popu lation exposure is the highest upper bound 95th per centile estimate, 0.014 pg/mL, from the adult blood donor biomonitoring study (Olsen et al., 2003c). 2.3. Derivation o f benchmark internal concentration (LBM ICi0) values 2.3.I. Dose metrics and the use o f serum PFOA concen tration data This risk characterization uses either: (1) measured serum PFOA concentration at presumed steady state; (2) pharmacokinetic estimates of steady state; or (3) 24-h mean serum PFOA concentration, as a measure that is associated with biological responses to PFOA and compared to representative general population serum PFOA concentrations. This method requires that PFOA be accurately measured in serum, and that the measured 368 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 serum PFOA concentration be related to response in toxicological studies. It also requires an understanding of the pharmacokinetics related to dosing that affect bioavailability and serum concentration. The assump tion is made that the dose of PFOA is related to serum PFOA concentration, which, in turn, is associated with response. Factors affecting the measurement of serum PFOA concentration include dosage form and fre quency, absorption rate constant, elimination rate con stant, time of sampling relative to dose administration, and the precision and variability in the analytical method (measurements of serum PFOA used in this characterization had a coefficient of variation of ap proximately 30%). If a non-metabolized compound is absorbed readily and poorly excreted, these factors have less influence on the value of measured serum concen tration than if it is readily absorbed and quickly ex creted. In the latter case, time of sampling relative to dose administration becomes critical, since the serum concentration of the compound changes rapidly. The striking differences in PFOA elimination rates between species, or sexes within species, are demonstrated in Table 7. These differences in PFOA elimination rate are largely overcome in estimating margins of exposure based on direct comparisons of serum concentrations at steady state or by using area-under-the-serum-concentration-versus-time curve (AUC, pg h/mL) to estimate steady-state or mean 24-h serum concentrations. For example, humans, who have a long elimination half-life (Burris et al., 2002), would exhibit de minimis change in their serum concentration over 24 h; thus, the 24-h mean serum concentration (AUC/24 h) represents steady state and is essentially the measured serum PFOA concen tration. In the case of adult female rats, which would have a rapidly changing profile of PFOA serum concentration due to essentially complete excretion of PFOA in a 24-h period (Table 7), AUC/24 h provides a time-weighted average serum concentration over the course of a dosing interval that can be used for com parison to human values. AUC, has been measured for male and female rats given single oral doses of PFOA (Kemper, 2003). The strong linear relationship between AUC and oral dose at the doses employed (0.1, 1, 5, and 25 mg/kg) allows interpolation of the value of AUC for any dose within the range 0.1-25 mg/kg (Fig. 1). AUC can be used to estimate steady state, be cause, in the theory of linear pharmacokinetics, the AUC from time zero to infinity for a single dose (AUC, pg h/mL) will equal the AUC for a dosage in terval at steady state, assuming that absorption and elimination rates remain constant from dose to dose (Wagner, 1975). Therefore, when dosing intervals are shorter than the elimination half-life, steady-state serum concentration may be estimated by AUC/24 h. The use of these concepts in estimating serum PFOA values for use in risk characterization is described below. Serum PFOA concentrations corresponding to a 10% change from normal or control were determined using USEPA National Center for Exposure Assessment software (BMDS version 1.3.1) and in general accor dance with a draft guidance document by USEPA (USEPA, 2000) on use of the benchmark dose. For most categorical data from toxicology studies, a 10% response level is fairly representative of the limits in which a change can be accurately determined. For continuous, normally distributed data, a shift in the distribution of 1.0 standard deviation represents approximately an extra 10% of the individual values being greater than approximately the 99th percentile or about an extra 10% less than approximately the 1st percentile of the Table 7 Reported values for elimination T1=2 o f perfluorooctanoate in various species Species Sex Dose form Observed T1/2 (days) Rat Rat Rat Rat Rat Rat Rat Rat Rat Rat Rat Rat Rat Dog Dog Monkey Monkey Retired workers Male Male Male Male Male Male Male Pregnant female Female Female Female Female Female Male Female Male Female Male Oral Inhalation Dermal Oral Oral i.p. i.v. Oral (GD 8-9) Oral Oral Oral i.p. i.v. i.v. i.v. i.v. i.v. Occupational exposure 5 5-7 5-7 9 (liver) 6-8 15 5.6 <0.5 <0.5 0.13-0.67 2.5 (liver) <1 0.08 20 and 23 8-13 21 33 1600i 1300 N 3 4-5/group 5/group 6 4 4 3 4 2 4 6 4 3 2 2 3 3 9 Reference Gibson and Johnson (1979) Kennedy et al. (1986) Kennedy (1985) Ylinen et al. (1990) Kemper (2003) Vanden Heuvel et al. (1991) Ohmori et al. (2003) Gibson and Johnson (1983) Gibson and Johnson (1983) Kemper (2003) Ylinen et al. (1990) Vanden Heuvel et al. (1991) Ohmori et al. (2003) Hanhijrvi et al. (1988) Hanhijrvi et al. (1988) Noker (2003) Noker (2003) Burris et al. (2002) J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 369 Fig. 1. Relationship of single oral dose of ammonium perfluorooctanoate (APFO) to serum perfluorooctanoate (PFOA) area-under-the-curve in male and female rats over a range of oral doses. These relationships are used in estimating mean serum concentrations of PFOA at the LBM D1o . The AUC corresponding to the LBM D10for a response in male or female rats is divided by 24 h to obtain a time-weighted average serum concentration over the daily dosing interval. For example, the dose corresponding to the LBM D10 for post-natal effects (22 mg/kg/day) is marked with an arrow. The corresponding AUC (^700 pg h/mL) was divided by 24 h to obtain the estimated LBMIC of 29 pg/mL. Data are based on Kemper (2003). distribution in controls. For continuous data, linear, polynomial, Hill, and power models were attempted, and those providing adequate goodness-of-fit, as deter mined by p > 0.05 and reasonable fit of modeled doseresponse curves in the range of the benchmark response on visual inspection, were used. In certain cases, elimi nation of the high-dose group was necessary to obtain adequate fit in the curve region in which the benchmark response level occurred. The resulting values from the modeling are referred to in this analysis as the Bench mark Internal Concentration for a 10% response (BMIC10), and the lower 95% CL of the BMIC10 (LBMIC10) was used as the point of departure (POD) for estimating MOE values. Serum concentrations representing the LBMIC10were determined by either of three methods. For the 13-week dietary study of APFO toxicity in male Sprague-Dawley rats (Palazzolo, 1993) and the six-month oral dosing study of APFO in cynomolgus monkeys (Butenhoff et al., 2002b), in which appropriate serum PFOA concentration data were available, the serum PFOA concentration was entered into the BMDS modelling program as the "dose," and this will be referred to as the "calculated" LBMIC10. For the cynomolgus monkey study, in order to derive LBMD10 and LBMIC10 values for body-weight change, serum PFOA concentrations at scheduled or unscheduled termination of dosing were used, along with any valid serum PFOA determinations within a two-week period prior to cessation of dosing. A second method was used for the two-generation reproduction study in SpragueDawley rats (Butenhoff et al., 2004), in which repre sentative serum PFOA concentration data were not available for all groups. In this case, the lower 95% CL of the benchmark administered dose for a 10% change (LBMD10) was determined, and the equations for AUC versus dose for male and female rats from Fig. 1were used to calculate AUC from the LBMD10. The resulting AUC was divided by 24 h to provide an average serum con centration over a 24-h dosing interval, and the result of this method will be referred to as the "estimated" LBMIC10. For example, a LBMD10 value of 22 mg/kg was used to calculate a corresponding estimated LBMIC10 of 29 pg PFOA/mL serum by calculating the AUC from the relationship in Fig. 1 (AUC = 31.5 x 22 mg/kg = 704 pg h/mL) and dividing by 24 h (704 pg h/mL/ 24 h = 29 pg/mL). The latter method provides an esti mated mean serum PFOA concentration over the course of a daily gavage-dosing interval that should approximate steady-state serum PFOA concentrations in adult male rats (see Section 2.1 above). In the case of adult female rats, which would not reach a steady-state serum PFOA concentration on daily gavage dosing due to rapid elimi nation of PFOA (Table 7), the latter method provides a mean 24-h serum PFOA concentration that is more meaningful than a value taken at a single time point on a rapid serum PFOA elimination curve that effectively reaches baseline within 24 h. For weanling rats, with the exception of one report in the literature (Kojo et al., 1986) that is lacking in detail, the elimination kinetics have not been described previously. However, recent work inves tigating the time course of development of sex differences in elimination of PFOA indicates that PFOA elimination in weanling rats is intermediate between adult male and female rats until four to five weeks of age, at which time the sex differences observed in adult rats become apparent (Han, 2003). Therefore, use of the adult female rat AUCversus-oral-dose relationship shown in Fig. 1 to estimate 24-h mean serum PFOA concentrations in F 1 male and female weanling rats between three and four weeks of age is warranted, if not conservative. Finally, serum PFOA 370 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 concentrations. The MOE was estimated by dividing the lowest LBMICi0 values (or, points of departure) for the chosen toxicological endpoints in Table 2 by human se rum PFOA concentrations considered to represent upper bound of the 95th percentile estimate of the general population serum PFOA concentration (0.014 pg/mL). Fig. 2. The relationship of serum perfluorooctanoate concentration to dose in male rats after dietary treatment with ammonium perflu orooctanoate (APFO) for 13 weeks. These data were used in calcu lating LBMIC10 for liver-to-brain-weight ratio and body-weight change in the 13-week dietary study (Palazzolo, 1993) and in esti mating the LBMIC10 from the LBM D10 for Leydig cell tumors (4.8 mg/kg/day, arrow) in the two-year dietary cancer bioassay dietary toxicology studies of ammonium perfluorooctanoate (APFO) in male rats (Sibinski et al., 1983). Data are derived from Palazzolo (1993). data were also not available for Leydig cell adenoma in cidence in rats from the two-year dietary cancer bioassay (Sibinski et al., 1983). In this case, the LBMD10 was de termined (USEPA, 1996, 1999, 2003c) after obtaining tumor incidences shown in Table 6 by the Poly-3 proce dure (Bailer and Portier, 1988) for calculating the effective lifetime number of rats at risk for cancer. The LBMIC10 was estimated using the serum PFOA concentration data from the subchronic study in rats (Palazzolo, 1993). Se rum PFOA concentration versus mean mg/kg/day intake of APFO from Table 4 were fit to a power curve (Fig. 2) to provide a means of interpolating the serum value corre sponding to the LBMD10 for Leydig cell tumors, and, thereby, estimating the LBMIC10. 2.4. Estimation o f margins o f exposure In this risk characterization, MOE values are calculated based on comparisons of serum PFOA 3. Results of calculated or estimated of LBMIC10 values 3.1. Post-natal developmental effects Benchmark doses (LBMD10) calculated for post-na tal developmental endpoints of F 1 pups in the twogeneration reproduction study (Butenhoff et al., 2004) are shown in Table 8. Because data on serum PFOA concentration in pups were not available from the study, the LBMIC10 correlating to the lowest LBMD10 for post-natal developmental effects (22 mg/kg/day based on F 1 female post-weaning mortality, plot for LBMD10 shown in Fig. 3) was calculated based on the relation ship of adult female rat AUC to administered dose, as explained in Section 2.3. Using this relationship, the value of AUC/24 h at a dose of 22 mg/kg/day was cal culated to be 29 pg/ml. The fact that adult F0 males had Fig. 3. Plot of BMDS-modeled dose (mg/kg/day)-response curve for post-weaning mortality in F 1 female rats (dichotomous multi-stage model, 10% response level). Table 8 Estimates of the BM D10, LBM D10, and LBMIC10 for post-natal effects in F 1 rat pups in a two-generation reproduction study with ammonium perfluorooctanoate (Butenhoff et al., 2004) Effect Model p valuea BM D10 (mg/kg/day) LBM D10 (mg/kg/day) LBMICiob (ig/m L) Days to preputial separation Linear 0.33 27 22 29 Post-lactational mortality in females Multistage 0.34 31 22 29 Post-lactational mortality in males Multistage 0.96 33 24 32 Days to vaginal patencyc Linear 0.001 41 30 40 Pre-weaning mortality (both sexes) Multistage 0.39 39 34 45 Day 22 pup weight (both sexes) Linear 0.21 97 44 59 aGoodness-of-fit p values greater than 0.05 indicate an adequate fit. bEstimated based on relationship of AUC to dose; e.g., the LBM D10 value of 22 mg/kg was used to calculate a corresponding estimated LBMIC10 of 29 pg PFOA/mL serum by calculating the AUC from the relationship in Fig. 1 (AUC = 32 x 22 mg/kg = 704 pg h/mL) and dividing by 24 h (704 pg h/mL/24 h = 29 pg/mL). cNone of the available models provided a good fit to the data. J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 371 Table 9 LBM D10 and LBMIC10 values for liver-weight-to-brain-weight ratio and for body-weight change in male rats and monkeys dosed with ammonium perfluorooctanoate Species/study LBM D10 values LBMIC10 values Model p valuea LBM D10 (mg/kg) Model p value LBMIC10 (pg/mL) Liver-weight-to-brain-weight ratio F0 rats/2-Genb Hill F0 rats/2-Gen Linear F1 rats/2-Gen Hill F1 rats/2-Gen Linear Rats/13-weeke Linear Monkeys/6-mo.f Body-weight changeh F0 rats/2-Gen F0 rats/2-Gen F1 rats/2-Gen Rats/13-week Linear Linear Polynomial Power1 Power Monkey/6-mo. Monkeys/6-mo. Power 0.30 0.07 0.25 0.17 0.18 0.01g 0.12 0.25 0.20 0.46 0.34 0.60 1.0 0.60 1.3 1.4 3.9 9.2 5.2 1.5 3.0 10 NAc NA NA NA Linear NA Linear NA NA NA Power NA Power Linear NA NA NA NA 0.23 NA 0.39 NA NA NA 0.42 NA 0.55 0.45 25d 42 25 54 34 58 23 380 220 63 88 130 64 60 aGoodness-of-fit p values greater than 0.05 indicate an adequate fit. bTwo-generation reproduction study (Butenhoff et al., 2004). cN ot applicable. The value o f the LBMIC was estimated based on the LBMD (see footnote d). dValues in italics are estimated based on relationship o f AU C to dose; e.g., the L BM Dio value o f 0.6 mg/kg was used to calculate a corresponding estimated LBMIC10o f 25 pg PFOA/mL serum by calculating the AUC from the relationship in Fig. 1 (AUC = 1000 x 0.6 mg/kg = 600 pg h/mL) and dividing by 24 h (600 pg h/mL/24 h = 25 pg/mL). eThirteen-week dietary study (Palazzolo, 1993). f Six-month oral dosing study in male cynomolgus monkeys (Butenhoff et al., 2002b). gEven though p < 0.05, this provides some reasonable level o f fit and is shown for comparative purposes. hBody-weight change from initiation o f dosing through termination o f dosing. i Unrestricted power model (power not restricted to p1). mean serum concentrations at termination of 51 9.3 and 45 13 pg/mL at 10 and 30 mg/kg/day, respectively (Butenhoff et al., 2004), suggests that a LBMIC10 based on measured adult male values would be 45-50 pg/mL. Therefore, it is conservative and appropriate to use the adult female-based LBMIC10 (29 pg/ml) for estimation of MOE values based on post-natal developmental ef fects. 3.2. Liver-weight increase 3.2.1. Rats The LBMD10 values and calculated or estimated (AUC/24 h) LBMIC10 for male rat liver-weight-tobrain-weight ratio increases are shown in Table 9. As can be seen, the Hill and linear models provided ade quate fits for LBMD10, and all resulting values (0.60 1.4) are tightly clustered within a factor of 2.4. There is also little difference between F 0 and F 1 males. The linear model provided an adequate fit for the LBMIC10 based on the serum PFOA concentration data from the Palazzolo (1993) study. Although not used for estimation of MOE values, LBMD10 and LBMIC10 values for ab solute liver-weight increase and liver-weight-to-bodyweight ratio were in the range of those derived from liver-weight-to-brain-weight ratios (data not shown). 3.2.2. Monkeys In the six-month oral-dosing study in male cyno- molgus monkeys (Butenhoff et al., 2002b), liver-weightto-brain-weight ratios were elevated at all APFO treatment levels relative to controls (Table 5); however, this parameter did not show an increase in response with increasing dose. Using values for dose, serum concen tration as related to dose, and liver-weight-to-brainweight ratio from Table 5, the calculated values of LBMD10 and LBMIC10 for liver-weight-to-brain-weight ratio are shown in Table 9. Fig. 4 shows the resulting plot that provided the LBMIC10 of 23 pg/mL used in MOE calculations. 3.3. Body-weight change 3.3.1. Rats LBMD10 and LBMIC10 values (calculated or esti mated) for body-weight change from the two-generation study (Butenhoff et al., 2004) F 0 and F 1 males, and the 13-week dietary study (Palazzolo, 1993) males are pre sented in Table 9. 3.3.2. Monkeys Body-weight change (decrease), a prominent effect no ted in male cynomolgus monkeys of the 30/20 mg/kg/day 372 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 Fig. 4. Plot of BMDS-modeled serum PFOA concentration (axis label ``dose," ig/m L) versus liver-weight-to-brain-weight ratio for male cynomolgus monkeys (linear continuous model, top dose-group data eliminated, one standard deviation from mean benchmark response, constant variance). Fig. 5. Plot of BMDS-modeled serum PFOA concentration (axis label "dose," ig/m L) versus body-weight change for male cynomolgus monkeys (linear continuous model, one standard deviation from mean benchmark response, constant variance). dose group in the six-month oral dosing study (Buten hoff et al., 2002b), produced calculated values of LBMDio and LBMICi0 presented in Table 9. The plot providing the LBMIC10 of 60 ig/mL that was used in estimating the MOE for body-weight change is shown in Fig. 5. 3.4. Leydig cell adenoma in rats Poly-3 (Bailer and Portier, 1988) adjusted incidence of Leydig cell adenoma in rats from the Sibinski et al. (1983) study are presented in Table 6. Fitting the multistage model to these age-adjusted incidence rates gave an esti mated LBMD10of 100 ppm in diet (4.8 mg APFO/kg/day) for a lifetime for Leydig cell tumors in male rats (Table 6 and Fig. 6). The latter mg/kg/day value is interpolated from the average estimated mg/kg/day dose of APFO over the two-year study in male Sprague-Dawley rats (1.3 mg/ kg/day for the 30 ppm dose group and 14 mg/kg/day for the 300 ppm dose group, Table 6). Using the equation of Fig. 6. Plot of BMDS-modeled dose (ppm APFO in diet)-response curve for Leydig cell adenoma in male rats (dichotomous multi-stage model, 10% response). the trendline in Fig. 2 (y = 46.1 x x06347) yielded a LBMICio of 125 ig PFOA/mL serum at 4.8 mg/kg/day (Table 6). The serum PFOA concentration data used in this analysis were obtained after dietary dosing for 14 weeks, and male rats in the 100 ppm dose group in the subchronic dietary study (Palazzolo, 1993) had received a mean of 6.5 mg/kg/day over the 14-week period. Since the LBMD10 for Leydig cell tumors of 100 ppm APFO in diet is near the geometric mean of the two dietary APFO dose groups in the cancer study (30 and 300 ppm), the geometric mean of the two conversion factors for ppm APFO to mg/kg/day APFO at 14 weeks in the cancer study could be used to estimate the mg/kg/day dose at 100 ppm after 14 weeks. Mean mg/kg/day doses at 30 and 300 ppm were 1.75 and 21.5 mg/kg/day, respec tively, after 14 weeks. Therefore, at the LBMD10 of 100 ppm, the estimated conversion factor would be 1ppm = (0.0717 x 0.0650)05 = 0.068 mg/kg/day. Hence, over the first 14 weeks in the cancer study, the average dose at the LBMD10 for Leydig cell tumors is estimated to be (100 x 0.068) = 6.8 mg/kg/day, which is quite comparable to the mg/kg/day dose from the subchronic study and corresponds to 156 ig PFOA/mL serum based on Fig. 2. In another approach, use of the relationship in Fig. 1 to estimate LBMIC10 at 4.8 mg/kg/day from AUC yields 204 ig/mL for the value of AUC/24 h. Thus, the use of the LBMIC10 value of 125 ig/m L based on 4.8 mg/kg/day is most conservative and will be used in estimating the MOE. 4. Characterization of risk 4.1. Points o f departure Table 10 presents the values of the LBMICio for post natal developmental effects in rats, liver-weight-tobrain-weight ratio increase in monkeys, body-weight change in monkeys, and tumorigenesis (Leydig cell) in rats that were used as POD to estimate MOE values J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 373 Table 10 Margin o f exposure values based on various LBMIC10 points o f departure and highest upper bound 95th percentile estimate o f general population serum PFOA concentrations Response (species) Source table Point o f departure LBMIQo (pg/mL) Margin of exposure1 Post-natal effects (rats) Liver-weight-to-brain-weight ratio0 (monkeys) Body-weight change (monkeys) Leydig cell tumors (rats) Table 8 Table 9 Table 9 Table 6 29b 23 60 125 2100 1600 4300 8900 aThe margin o f exposure is calculated by dividing the LBMIC10 (pg/mL) by the general population serum [PFOA] representing the upper 95% confidence limit o f the estimate o f the 95th percentile general population serum [PFOA] (0.014 pg/mL). Margins o f exposure based on the upper bound o f the geometric mean general population serum [PFOA] (0.005 pg/mL) are approximately three times higher. bThe serum [PFOA] in post-weaning rat pups was estimated conservatively based on adult-female rat AU C at the LBM D10 value o f 22 mg/kg/day for post-natal effects using the relationship o f AUC to administered oral dose from Fig. 1. Results from studies currently in progress support the premise that this is a conservative estimate for weanling rat pups (Han, 2003; Mylchreest, 2003). Availability o f data in the future may require adjustment o f this estimate o f the MOE. cLiver-weight increase is not necessarily reflective o f an adverse effect, as this is a normal adaptive response. This endpoint was used as a sensitive indication o f biological response. based on serum PFOA concentration. The POD LBMIC10 values ranged from 23 pg/mL (liver-weight-tobrain-weight ratio increase in monkeys) to 125 pg/mL (Leydig cell adenoma in rats). 4.2. Margins o f exposure The results of dividing POD LBMIC10 values by the upper-bound estimated 95th percentile general popula tion serum concentration (0.014 pg/mL) to estimate MOE values are presented in Table 10. These MOE values varied from 1600 for increased liver-weight-tobrain-weight ratio in monkeys to 8900 for Leydig cell adenoma in rats. 4.3. Analysis o f uncertainty This subsection discusses factors that tend to reduce uncertainty as well as residual uncertainty in this risk characterization. Current methods for qualitatively as signing uncertainty factors in a deterministic approach to assessing risk have been recently reviewed by Kalberlah et al. (2003). Most methods currently used to assign uncertainty include factors for intraspecies and interspecies extrapolation. Subdividing the latter two uncertainty factors into subfactors for toxicokinetics and toxicodynamics has gained acceptance. A third area of uncertainty comes from differences in length of ex posure (chronicity). 4.3.1. Toxicokinetic factors 4.3.1.1. Interspecies toxicokinetic comparisons. A num ber of factors reduce uncertainty with respect to interpsecies differences in toxicokinetics. A primary factor is that PFOA is not metabolized; therefore, species differ ences in metabolic handling do not need to be accounted for. A second factor is the availability of sound PFOA toxicokinetic data from multiple studies and species, including pregnant rats, weanling rats, rodents, dogs, monkeys, and humans (Table 7; Goecke et al., 1992; Kuslikis et al., 1992; Kerstner-Wood et al., 2003). These studies provide a reasonable understanding of absorp tion, distribution, and elimination of PFOA across species. With respect to residual uncertainty in toxic okinetics between species, the notable sex and species differences presented in Table 7 can be overcome, in part, through the use of direct serum PFOA compari sons or comparisons with estimated serum PFOA con centrations based on pharmacokinetic factors such as AUC. The volume of distribution in male rats and male and female cynomolgus monkeys is quite similar, and suggests that PFOA is distributed primarily in extra cellular spaces (Kemper, 2003; Noker, 2003). In addi tion, PFOA has been shown to have similar plasma binding characteristics in rats, monkeys, and humans (Kerstner-Wood et al., 2003). A third factor that is of considerable value in reducing uncertainty, when esti mating MOE values based on comparisons of serum PFOA concentrations associated with a level of response in toxicological studies with general population serum PFOA concentration, is the availability of human serum PFOA concentration data from large cohorts of the general population that allow for analysis of differences by age and gender and include children, adults, and the elderly. Use of the upper bound of the highest 95th percentile estimated general population serum concen tration for comparison with POD LBMIC10 values from toxicological studies narrows the MOE, thus accounting for any uncertainty relative to the variability in the distribution of human serum PFOA concentrations. The MOE estimates are based on the assumption of steady-state serum PFOA concentrations, with the ex ception of the female rat, for which 24-h average serum PFOA concentrations based on the relationship of dose to AUC were used. A near steady state is likely in the general population because of the prolonged elimination half-life (Burris et al., 2002) and minimal exposure (plus 374 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 the tight distribution of PFOA serum concentrations from individual sampling of children, adults, and the elderly). During repeated daily dosing or dietary intake in toxicology studies, male rats and monkeys appear to reach steady state within approximately one month (Butenhoff et al., 2002b; Kemper, 2003); therefore, un certainty around the use of serum level is further mini mized. Female cynomolgus monkeys have a similar, perhaps slightly longer (approximately 30 versus 20 days), elimination half-life as compared to males (Noker, 2003). In the case of the female rat, a steadystate condition cannot be reached on daily dosing due to the rapid elimination of PFOA (see discussion in text). As a result, the authors believe that mean 24-h serum PFOA concentration (AUC divided by 24 h) is an ap propriate comparative measure. 4.3.I.2. Intraspecies toxicokinetic comparisons. Individ ual differences in metabolism do not exist. The general population monitoring data available for PFOA also decrease uncertainty related to the variability in expo sure, since serum PFOA concentrations have been characterized in a group of children, in adult American Red Cross blood donors, and in an elderly population (Table 1). Group sizes were large and covered a rea sonably representative geographical distribution and age distribution. The fact that these serum measurements are tightly distributed and do not show major differences between age or gender groups (Table 1) reduces the chance that large portions of the population may not be adequately represented in a risk characterization. The tight distribution of general population serum PFOA concentrations (Table 1), particularly across age groups and between sexes, is not consistent with initial expec tations. If the elimination half-life in humans is indeed in the range of that estimated by Burris et al. (2002), then steady-state serum PFOA concentrations would not be expected until after 5-25 years of exposure or longer. However, PFOA concentrations in general population serum do not increase with age. Therefore, the estimated serum elimination rate provided by Burris et al. (2002) for retired workers may not be representative of the elimination rate for the general population with much lower serum PFOA concentrations. The external expo sure and pharmacokinetic factors that influence the observations from the general population biomonitoring studies of Olsen et al. (2003c, 2004a,b) remain to be discovered. Once again, these matters point to the utility of risk characterization based on direct comparison of serum concentration, which has the effect of reducing overall uncertainty. Another fact that aids in reducing uncertainty in in traspecies toxicokinetic factors is that medical moni toring of exposed workers has included measurements of serum PFOA concentration for a quarter century (3M Company, 2003a,b; Gilliland, 1992; Gilliland and Mandel, 1996; Olsen et al., 1998, 2000; Ubel et al., 1980). Ubel et al. (1980) originally reported that serum total organic fluorine concentrations (assumed to be due predominantly to the presence of PFOA) among 3M Cottage Grove, Minnesota fluorochemical workers ranged from 1 to 71 pg/mL. Serum PFOA concentra tions during the 1990s have remained comparable to those initially estimated by Ubel et al. (1980) with means (range in parentheses) of 5.0 pg/mL (0-80), 6.8 pg/mL (0.0-114), and 6.4 pg/mL (0.1-81) in 1993, 1995, and 1997, respectively. Serum PFOA concentrations of other 3M fluorochemical production workers in Decatur, Al abama, and Antwerp, Belgium have averaged 1-2 pg/ mL with highest concentrations reported to be 13 pg/mL (Olsen et al., 2003a). In estimating weanling rat serum PFOA concentra tions, the relationship between dose and AUC (Fig. 1) for adult female rats was exploited. Based on the study by Kemper (2003), the latter relationship is linear over a broad range of doses. The adult female serum concen tration associated with the LBMDio for post-natal ef fects can be calculated from this relationship. This was accomplished by taking the value of the AUC at the LBMD10 (pg h/mL) and dividing this by 24 h to provide an average serum concentration. Based on results from a recent study sponsored by 3M and DuPont (Han, 2003) the 24-h average adult female serum concentration may underestimate the actual concentration in weanling rats. The study by Han (2003) demonstrates that the elimi nation of PFOA in weanling rats is intermediate be tween the elimination in adult males and females until sometime between four and five weeks of age, when the hormonally regulated sex differences in PFOA elimina tion became apparent. In the latter study, 24 h after a single oral dose (10 mg/kg), female serum PFOA con centrations were 2.4-fold lower at five weeks of age than at four weeks of age and were not further affected through study termination at eight weeks of age. Serum PFOA concentrations in males were 2.7-fold higher than females at four weeks of age, and five-week-old males were 5.4-fold higher than four-week-old males. This recent finding is consistent with the suggestion by Kudo et al. (2002) that sexual hormone regulation of the ex pression of certain organic anion transporters in kidney (OAT2, OAT3, and oatp1) may account for sex differ ences in PFOA elimination in rats. They found OAT2 to be more highly expressed in female rat kidney and subject to up-regulation by estradiol. Buist et al. (2002) examined differences in expression of organic anion transporter proteins during post-natal development in the rat. These researchers confirmed the much greater expression of OAT2 in female rat kidney as compared to male rats and showed that OAT2 does not increase in expression during development through post-natal day 45 in the male but does increase between post-natal days 35 and 40 in females. Based on these recent studies, it is J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 375 likely that male and female weanling rat pups in the first days after weaning (i.e., between three and four weeks of age), where a statistically significant increase in mor tality was observed, have lower PFOA excretion rates when compared to adult female rats. Based on another recent study co-sponsored by 3M and DuPont (Mylchreest, 2003), body burdens of PFOA from gestational and lactational exposure existed prior to the initiation of dosing at weaning in the two-generation reproduction study reported by Butenhoff et al. (2004). Therefore, the estimated adult female 24-h mean serum PFOA con centration for the LBMICi0 for post-natal develop mental effects is believed to provide a realistic and conservative estimation of serum PFOA concentrations for the weanling rats. In male rats, dietary dosing with APFO produced higher serum PFOA concentrations at a calculated mg/ kg/day dose level than did gavage dosing, as is evidenced on comparison of steady-state serum PFOA concentra tion data from the two-generation study by gavage (Butenhoff et al., 2004) with the 13-week subchronic dietary study (Palazzolo, 1993). This may be due to potentially lower absorption with a bolus dose as com pared to lower level continuous dietary intake. Another possible explanation for this difference may relate to the observed elevation of estradiol in male rats on treatment with APFO (Biegel et al., 2001; Liu et al., 1996b), which may be more pronounced in gavage dosing due to po tentially higher Cmax values. This could lead to a higher rate of increased urinary excretion in gavage dosing due to greater up-regulation of urinary transport systems by elevated estradiol in male rats, as observed by Kudo et al. (2002). Considering potential male/female differences in tox icokinetics within primate species, male and female rhesus monkeys had similar serum PFOA concentra tions after dosing for 90 days at 3 and 10 mg/kg/day (Griffith and Long, 1980). In cynomolgus monkeys, a sex difference in elimination rate of less than twofold has been noted after a single i.v. dose of potassium PFOA, with females having the lower mean elimination rate (Noker, 2003). 4.3.2. Toxicodynamic factors The toxicodynamic response in experimental studies can be related to serum PFOA concentration, and, in some cases, target tissue dose. Therefore, it is possible to gain insight into the variability of some toxicodynamic responses across species and within a species. This is true not only for the species used in experimental studies, but also for humans. 4.3.2.I. Interspecies toxicodynamic comparisons. This risk characterization benefits from the availability of a large number of studies covering most toxicological endpoints of interest. The fact that non-human primate toxicology studies are available as well as epidemiolog ical and medical monitoring studies of PFOA-exposed workers is a significant factor in reducing uncertainty with regard to extrapolation of responses from studies with test species to humans. A number of the toxico logical studies as well as the worker-health studies have measured serum PFOA concentrations that can be as sociated with observations from the studies. For studies where serum PFOA concentration data are not avail able, estimates of serum PFOA concentration can be made using established pharmacokinetic factors. Therefore, it was possible in this risk characterization to relate serum PFOA concentration to selected responses across species. In inspecting the results of modeled doseresponse and serum-PFOA-concentration-response re lationships for liver-weight increase and body-weight change from Table 9, the LBMICio values for male rats and monkeys are reasonably comparable, a fact that supports the comparison of serum PFOA concentration associated with response across species. With respect to liver responses, mitochondrial pro liferation has been observed in rats and monkeys (Berthiaume and Wallace, 2002; Butenhoff et al., 2002b); however, only rats have shown increased PPAR-a agonism (peroxisome proliferation) after treatment with PFOA (Berthiaume and Wallace, 2002; Biegel et al., 200i; Palazzolo, i993), a fact that is consistent with primates being generally non-responsive to PPAR-a agonists (Ashby et al., 1994; Bentley et al., 1993; Cattley et al., 1998). Medical monitoring of human workers exposed to PFOA has not shown associations with liver function abnormalities (liver enzymes in serum) or other measured endpoints at serum PFOA concentrations within an order of magnitude of the LBMIC10 for bodyweight and liver-weight effects in male monkeys (Olsen et al., 2000). The relevance to humans of the observed tumors in the two chronic dietary studies (Biegel et al., 2001; Sibinski et al., 1983) is uncertain. The hepatocellular tumors observed by Biegel et al. (2001) are likely to be related to PPAR-a agonism; therefore, they likely are not relevant to humans (Ashby et al., 1994; Bentley et al., 1993; Cattley et al., 1998). The Leydig cell tumors may result from hormonal changes brought about by induction of aromatase (Biegel et al., 1995; Liu et al., 1996a,b), and this proposed mechanism, in addition to being nonlinear, would be of questionable relevance to humans (Clegg et al., 1997; Cook et al., 1999). The in cidence of pancreatic acinar cell adenomas in the Biegel et al. (2001) study was 0/80, 1/79 (1.3%), and 7/76 (9.2%) in control, control pair-fed, and 300 ppm PFOA groups, respectively. A higher incidence of pancreatic acinar cell adenoma was also observed with the potent PPAR-a agonist, WY-14,643, in the same study (Biegel et al., 2001). The mechanism of PFOA-induced pancreatic acinar tumors in rats remains to be elucidated. Most 376 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 human pancreatic cancers are of ductal origin, and current understanding of the mechanisms underlying human pancreatic cancers suggest that mutations in oncogenes (predominantly K-ras) and genes coding for certain tumor suppressor factors are involved in most human pancreatic cancers (Anderson et al., 1996; Fernandez-Zapico et al., 2003; Li and Jiao, 2003; Moore et al., 2003; Schneider and Schmid, 2003; Urrutia, 2002). PFOA has not shown mutagenic or clastogenic activity in a variety of standard assay systems, and is unlikely to be a complete carcinogen. A statistically significant in crease in acinar cell adenoma was observed in one of two rat two-year studies at a dose of 300 ppm in diet, although, acinar cell hyperplasia was evident in both studies (Frame and McConnell, 2003). If the assumption is made that the PFOA-induced acinar cell tumors in rats in the Biegel et al. (2001) study are likely to have been the result of mechanisms that involve epigenetic or proliferative mechanisms as op posed to direct mutations, it is likely that the dose-re sponse curve is nonlinear; therefore, it would be appropriate to consider benchmark-dose methodology in risk characterization. When compared to the ageadjusted Leydig cell adenoma incidence in the Sibinski et al. (1983) study (0/48, 2/48, and 7/48 for the control, 30 ppm, and 300 ppm dose groups, respectively) for which an LBMIC10 has been estimated, it becomes evi dent that the LBMIC10 for Leydig cell tumor incidence could be expected to be lower than that for pancreatic acinar cell tumors in the study by Biegel et al. (2001). Therefore, Leydig cell tumors were used in this risk characterization to represent nonlinear cancer risk. 4.3.2.2. Intraspecies toxicodynamic comparisons. Medical monitoring and epidemiological studies among 3M Company fluorochemical workers in Cottage Grove, Minnesota, that were engaged in PFOA production and processing have not found associations of PFOA expo sure with altered health status, including clinical chem istry and hormonal abnormalities, and no statistically significant increases in standardized mortality ratios (SMR) were found for total cancer (SMR = 0.9, 95% CI 0.7-1.1), liver cancer (SMR = 0.6, 95% CI 0.1-3.3), or pancreatic cancer (SMR = 1.4, 95% CI 0.5-3.1) (3M Company, 2003a,b; Alexander, 2001; Gilliland, 1992; Gilliland and Mandel, 1996; NIOSH, 2001; Olsen et al., 1998, 2000, 2003a). There was one death attributed to testicular cancer (approximately 0.5 expected) among these fluorochemical production workers (Alexander, 2001). The low case-fatality rate for testicular cancer does not allow for a straight-forward interpretation of results from an occupational cohort mortality study. These epidemiological data provide a level of comfort in characterizing health risk of the population. The lack of observed effect at the higher serum PFOA concentra tions experienced by workers reduces uncertainty in considering the toxicodynamic response in non-occu pational populations with serum PFOA concentration levels two-to-three orders of magnitude lower. In regard to male/female differences, differences in response may be due, in part, to toxicokinetic differ ences. For example, in the rhesus monkey study re ported by Griffith and Long (1980), there were no obvious differences between the response of males and females; although, the numbers per dose group (two per sex) limit interpretation. This is consistent with serum PFOA concentrations being similar. In contrast to monkeys, adult male rats are notably more responsive to body-weight and liver-weight effects of PFOA than fe males at similar administered doses. The striking dif ferences in elimination rate between male and female rats (Table 7) may partially explain this apparent toxicodynamic difference. The differences in tumor outcome between the twoyear dietary studies with APFO in rats reported by Sibinski et al. (1983) and Biegel et al. (2001) is not un derstood. Both studies included a 300 ppm APFO die tary dose, and compound consumption over two years of dosing averaged 13.9 and 13.6 mg/kg/day in the Sib inski et al. (1983) and Biegel et al. (2001) studies, re spectively. Biegel et al. (2001) report purity of the sample to be 98-100%, and mixed the sample with Certified Rodent Diet #5002 (PMI Feeds, Inc.) in a Hobart mixer at high speed for 6 min. Sibinski et al. (1983) report sample purity to be 97.6-98.4% mixed C8 isomers, and the sample was determined to be 79% lin ear, with 9% terminal branching, and 12% backbone branching. The latter sample was mixed with Certified Purina Laboratory Chow (Ralston Purina, St. Louis, Missouri). Three possibilities arise that may explain the toxicodynamic differences seen in these two studies. These are: (1) possible differences in amount of branched vs. linear APFO; (2) possible influences of base diet; and (3) genetic drift in Sprague-Dawley rats over time. Other possibilities may also be raised, but it is not possible at this time to explain why hepatocellular ade noma and pancreatic acinar cell adenoma were seen by Biegel et al. (2001) but not by Sibinski et al. (1983). 4.3.3. Factors reducing uncertainty related to chronicity o f exposure The finding of PFOA in the serum of children, adults, and a group of elderly combined with the observed long elimination half-life suggests lifetime internal presence of PFOA. Several factors mitigate the degree of uncer tainty that is related in applying experimental study re sults to humans. First, experimental studies have been conducted that cover all periods of development and lifetime exposure in rats. These include the two cancer studies (Biegel et al., 2001; Sibinski et al., 1983), devel opmental toxicity studies in rats and rabbits (Gortner, 1981; Gortner, 1982; Staples et al., 1984), and the two- J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 377 generation reproduction study in rats (Butenhoff et al., 2004). A second factor that addresses uncertainty re lated to chronic exposure is the availability of epidemi ological and medical monitoring studies in exposed workers over several decades, where exposures on a se rum PFOA concentration basis have been significantly higher than those of the general population (3M Com pany, 2003a,b; Alexander, 2001; Gilliland, 1992; Gilli land and Mandel, 1996; NIOSH, 2001; Olsen et al., 1998; Olsen et al., 2000; Olsen et al., 2003a; Ubel et al., 1980). These worker studies have not found consistent health effect associations related to PFOA exposure. 5. Discussion 5.1. Approach The approach used in this risk characterization has the advantage of deriving MOE values using multiple biological responses in rodents and monkeys by com parison of concentrations of PFOA in serum. In addi tion, the use of LBMICio values as PODs has an advantage over the use of a NOAEL or LOAEL in that the benchmarked values represent a defined level of excess response (risk). This provides a better view of species differences in response by normalizing response at a specified level. Therefore, unlike the NOAEL or LOAEL from a study, the benchmark value is directly related to a given level of response, or risk, and can be used in probabilistic risk assessment (Gaylor and Kodell, 2002). The benchmark response level (BMR) that is used to calculate the benchmark dose may vary de pending on the endpoint that is being benchmarked. For most categorical data from toxicology studies, a 10% response level (BMD10) is fairly representative of the limits in which a change can be accurately determined. For continuous, normally distributed data, a shift in the distribution of 1.0 standard deviation represents ap proximately an extra 10% of the individual values being greater than near the 99th percentile or about an extra 10% less than near the 1st percentile of the distribution in controls. To be conservative, the lower 95% CL of the BMIC (LBMIC10) has been used. Uncertainty related to variations in response within and between species is further reduced by the fact that PFOA is not metabolized, and the pharmacokinetics of PFOA have been investigated in rodents and pri mates. The use of serum PFOA concentration also has advantages in that internal dose is more directly re lated to biological response (toxicodynamics), and the influence of rates of absorption and elimination (tox icokinetics) on response when using external (admin istered) dose are minimized. When all of the above advantages are combined, the result is reduction in uncertainty. Serum PFOA concentrations measured in children (Olsen et al., 2004a), adults (Olsen et al., 2003c), and the elderly (Olsen et al., 2004b) represent exposure from all sources. Age and sex of the individual human subjects was known; however, intensity and duration of exposure cannot be known. That said, the distribution of serum PFOA concentrations is comprised of data from several hundred individuals for adults and children, and over 200 for the elderly. In the case of adults and children, the samples are from a representative geographical cross section of the United States. Because these samples represent a ``snap-shot'' of serum PFOA concentration at a point in time for these individuals, the distribution will account for the variation in intensity and duration of exposure among individuals. As the data show, the distribution is remarkably tight, adding confidence to the suitability of our methodology for this risk charac terization. It could be argued that comparisons of PFOA con centrations in liver tissue, a primary target of PFOA toxicity, would be more meaningful for risk character ization. As stated in the introduction, Olsen et al. (2003b) examined a total of 30 human donor livers for the presence of PFOA. All donor livers were below the lower limit of quantitation for at least one of two analyses per sample except for one liver that had an average of 0.047 pg/g. Although many serum PFOA analyses in paired samples from this donor population were also below the limit of quantitation, there were more serum values that were quantifiable than liver values. Concentrations of PFOA in liver measured in toxicology studies in rats (Griffith and Long, 1980) and monkeys (Butenhoff et al., 2002b) have been comparable to serum concentrations. Also, analysis of pharmaco kinetic data obtained during the six-month oral toxicity study in male cynomolgus monkeys (Butenhoff et al., 2002b) does not suggest that PFOA is eliminated from liver to a lesser extent than it is eliminated from serum (Butenhoff et al., manuscript in preparation). Therefore, the authors believe that serum PFOA concentration can be correlated with effects that involve liver tissue. 5.2. Points o f departure A brief discussion on the choice of points of depar ture is in order. Post-natal developmental effects were used, specifically, the LBMD10 and LBMIC10 values associated with post-weaning mortality. This POD rep resents an adverse outcome in rats that has the greatest meaning for risk characterization. Liver-weight increase measures were also used as POD because this is a sen sitive response in male rats and monkeys. It must be emphasized that liver-weight increase does not neces sarily represent an adverse effect, as it is typically an adaptive response to dosing; thus, use of liver-weight increase as a point of departure is believed to be 378 J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 conservative. Liver injury does occur under higher dosing conditions, and liver is the primary target organ. The same thoughts can be expressed for body-weight change, a response that is not necessarily adverse. With PFOA, the threshold for body-weight change is higher than that for liver-weight increase, and body-weight change is closer on a continuum to the occurrence of adverse effects. As such, it is also a conservative end point for a point of departure, but has more relevance to potential adverse effects because it may represent effects on appetite or metabolism. Male rats and monkeys re spond with both liver-weight increase and body-weight change, with benchmark doses and internal concentra tions that are relatively similar. Finally, a benchmark dose was developed for Leydig cell tumors in the chronic dietary study of Sibinski et al. (1983). Leydig cell tumors are rare in humans (Anderson et al., 1996), and Biegel et al. (2001) suggest that the mode of action for APFOinduced Leydig cell adenoma may be a sustained in crease in estradiol after induction of aromatase. Due to the fact that APFO was administered at a single treat ment level (300 ppm in diet), dose-response data over a range of doses were not available for the liver and pancreatic acinar cell tumors observed in the study by Biegel et al. (2001). The weight of evidence pertaining to PPARa agonists would indicate that it is unlikely that the liver tumors are relevant to humans (Ashby et al., 1994; Bentley et al., 1993; Cattley et al., 1998). The pancreatic tumors observed in the Biegel et al. (2001) study were likely the result of epigenetic and/or prolif erative mechanisms, because pancreatic acinar cell pro liferation was elevated in 300-ppm-treated rats when measured at 15, 18, and 21 months. Furthermore, PFOA is not known to cause mutations or chromo somal aberrations. Therefore, PFOA is not likely to be a complete pancreatic carcinogen in humans. It would follow that a nonlinear model would be appropriate in assessing cancer risk from PFOA. In calculating benchmark dose and benchmark in ternal concentration, all models were explored, and, if the study data fit more than one model adequately, the results of calculations for multiple models were dis played. In choosing points of departure, the lowest LBMIC10 value for the endpoint was employed. The small number of animals in the monkey study (Buten hoff et al., 2002b) should be taken into consideration, and, given the somewhat similar responses to liverweight increase and body-weight change, use of the lower value regardless of species seems appropriate. 5.3. Context on the margins o f exposure Based on the discussion above, there are justifiable reasons to reduce the default uncertainty factors of 10 for intraspecies and interspecies uncertainty as well as chronicity of exposure in the traditional approach to uncertainty analysis. Among the most significant factors discussed above are the lack of metabolism of PFOA, the extensive experimental database, human biomoni toring data, occupational epidemiological studies, medical monitoring of workers, and the ability to relate serum PFOA concentration to effect. Because there are reasons to suggest reductions in the default uncertainty factors, it is reasonable to conclude that the lowest MOE determined in this risk characterization (1600) represents a substantial level of protection based on the current norms (e.g., European Commission, 2002). MOEs based on the geometric means of population se rum PFOA determinations would be approximately 2-3 times higher than those based on the upper bound of the 95th percentile estimated serum PFOA concentration. If the highest measured individual serum PFOA concen tration of 56 pg/mL from the biomonitoring studies of Olsen et al. (2003c, 2004a,b) is considered, MOE values would still indicate substantial protection. 6. Conclusion This risk characterization used responses that included postnatal developmental effects in rats, liverweight-to-brain-weight ratio increase in rats and mon keys, body-weight change in rats and monkeys, and increased incidence of Leydig cell adenoma. The upper bound of the highest 95th percentile estimated general population serum PFOA concentration (0.014 pg PFOA/mL) that occurred in three biomonitoring studies of United States general populations was used to rep resent human exposure. The use of serum PFOA con centration metrics in calculating MOE values reduced uncertainty in the risk characterization. Using ap proaches that relate serum PFOA concentration to re sponse, MOE values based on the upper bound 95th percentile estimated population serum PFOA concen tration were large, ranging from 1600 (liver-weight in crease) to 8900 (Leydig cell adenoma). These MOE values represent substantial protection of children, adults, and the elderly in the general population. References 3M Company, 2003a. Final Report, Assessment o f lipid, hepatic and thyroid function in relation to an occupational biologic limit value for perfluorooctanoic acid, 3M Medical Department, June 9, 2003. EPA Docket No. AR226-1351. 3M Company, 2003b. Final Report, Descriptive analysis o f serum fluorochemical concentrations from Cottage Grove employee participants o f the 2002 Medical Surveillance Program, 3M Medical Department, June 17, 2003. USEPA Public Docket No. AR-226-1352. 3M Company, 2003c. Environmental and health assessment of perfluorooctanesulfonate and its salts. St. Paul, Minnesota, August, 2003. USEPA Public Docket AR-226-1486. J.L. Butenhoff et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 379 Alexander, B., 2001. Mortality study o f workers employed at the 3M Cottage Grove facility. University o f Minnesota, Minneapolis, Minnesota. USEPA Public Docket AR-226-1030.a018. Anderson, K.E., Potter, J.D., Mack, T.M., 1996. Pancreatic cancer. In: Schottenfeld, O., Fraumeni, J. (Eds.), Cancer Epidemiology and Prevention. Oxford University Press, New York, pp. 725-771. Ashby, J., Brady, A.M ., Elcombe, C.R., Elliott, B.M., Ishmael, J., Odum, J., Tugwood, J.D., Kettle, S., Purchase, I.F.H., 1994. Mechanistically based human hazard assessment o f peroxisome proliferator-induced hepatocarcinogenesis. Human Exp. Toxicol. 13 (Suppl. 2), 1-117. Bailer, A.J., Portier, C.J., 1988. Effects o f treatment-induced mortality and tumor-induced mortality on tests for carcinogenicity in small samples. Biometrics 44, 417-431. Bentley, P.I., Calder, I., Elcombe, C.R., Grasso, P., Stringer, D., Wiegand, H.-J., 1993. Hepatic peroxisome proliferation in rodents and its significance for humans. Food Chem. Toxicol. 31, 857-907. Berthiaume, J., Wallace, K.B., 2002. Perfluorooctanoate, perfluorooctane sulfonate, and W-ethyl-perfluorooctanesulfonamido ethanol; peroxisome proliferation and mitochondrial biogenenesis. Toxicol. Lett. 129, 23-32. Biegel, L.B., Liu, R.C.M., Hurtt, M.E., Cook, J.C., 1995. Effects of ammonium perfluoroctanoate on Leydig cell function: in vitro, in vivo, and ex vivo studies. Toxicol. Appl. Toxicol. 134, 18-25. Biegel, L.B., Hurtt, M.E., Frame, S.R., O'Connor, J.C., Cook, J.C., 2001. Mechanisms o f extrahepatic tumor induction by peroxisome proliferators in male CD rats. Toxicol. Sci. 60, 44-55. Buist, S.C.N., Cherrington, N.J., Choudhuri, S., Hartley, D.P., Klassen, C.D., 2002. Gender-specific and developmental influences on the expression o f rat organic anion transporters. J. Exp. Pharmacol. Ther. 301, 145-151. Burris, J.M., Lundberg, J.K., Olsen, G.W., Simpson, C., Mandel, J., 2002. Interim Report N o. 2, Determination o f serum half-lives of several fluorochemicals. St. Paul (M N), 3M Company. US EPA Public Docket AR-226-1086. Butenhoff, J.L., Kennedy, G.L., Murphy, S.R., O'Connor, J.C., Olsen, G.W., 2002a. Genotoxicity, carcinogenicity, developmental effects, and reproductive effects o f perfluorooctanoate: a perspective from available animal and human studies. Association o f Plastics Manufacturers o f Europe and the Society o f the Plastics Industry. Submitted to USEPA docket AR-226, December 19th, 2002. Butenhoff, J.L., Costa, G., Elcombe, C., Farrar, D., Hansen, K.J., Iwai, H., Jung, R., Kennedy, G.L., Lieder, P., Olsen, G.W., Thomford, P., 2002b. Toxicity o f ammonium perfluorooctanoate (APFO) in male cynomolgus monkeys after oral dosing for six months. Toxicol. Sci. 69, 244-257. Butenhoff, J.L., Kennedy, G.L., Frame, S.R., O'Connor, J.C., York, R.G., 2004. The reproductive toxicity o f ammonium perfluorooctanoate. Toxicology 196, 95-116. Cattley, R.C., DeLuca, J., Elcombe, C., Fenner-Crisp, P., Lake, B.G., Marsman, D.S., Pastoor, T.A., Popp, J.A., Robinson, D.E., Schwetz, B., Tugwood, J., Wahli, W., 1998. D o peroxisome proliferating compounds pose a hepatocarcinogenic hazard to humans? Regul. Toxicol. Pharmacol. 27, 47-60. Clegg, E.D., Cook, J.C., Chapin, R.E., Foster, P.D., Daston, G.P., 1997. Leydig cell hyperplasia and adenoma formation: mechanisms and relevance to humans. Reprod. Toxicol. 11, 107-121. Cook, J.C., Murray, S.M., Frame, S.R., Hurtt, M.E., 1992. Induction o f Leydig cell adenomas by ammonium perfluorooctanoate: a possible endocrine-related mechanism. Toxicol. Appl. Pharmacol. 113, 209-217. Cook, J.C., Klinefelter, G.R., Hardisty, J.F., Sharpe, R.M ., Foster, P.M .D., 1999. Rodent Leyding cell tumorigenesis: a review o f the physiology, pathology, mechanisms, and relevance to humans. Crit. Rev. Toxicol. 29, 169-261. European Commission, 2002. First report on the harmonisation o f risk assessment procedures. Part 1: The report o f the Scientific Steering Committee's Working Group on harmonisation o f risk assessment procedures in the Scientific Committees advising European Com mission in the area o f human and environmental health, 26-27 October. Fernandez-Zapico, M.E., Bramati, P.S., Zakaria, S., Kaczynski, J.A., Urrutia, R., 2003. Fundementals o f transcription factors and their impact on pancreatic development and cancer. Pancreatology 3, 276-283. Feron, V.J., de Groot, A.P., Spanjers, M.T., Til, H.P., 1973. An evaluation o f the criterion organ weight under conditions o f growth retardation. Food Cosmet. Toxicol. 11, 85-94. Frame, S.R., McConnell, E.E., 2003. Review o f proliferative lesions of the exocrine pancreas in two chronic feeding studies in rats with ammonium perfluorooctanoate. Internal DuPont Project Identifi cation DuPont-13788 dated October 16, 2003. USEPA Public Docket AR-226. Gaylor, D.W., Kodell, R.L., 2002. A procedure for developing riskbased reference doses. Regul. Toxicol. Pharmacol. 35, 137-141. Gaylor, D.W., Moore, J.A., Butenhoff, J.L., 2003. The concept of benchmark internal concentration (BMIC) in risk assessment (abstract). Toxicol. Sci. 72 (S-1), 140-141 (Supplement, abstract #681). Gibson, S.J., Johnson, J.D., 1979. Absorption o f FC-143-14C in rats after a single oral dose. Riker Laboratories, Inc., Subsidiary o f 3M, St. Paul, M N, USEPA Public Docket AR-226-0455. Gibson, S.J., Johnson, J.D., 1983. Extent and route o f excretion of total carbon-14 in pregnant rats after a single oral dose of ammonium 14C-perfluorooctanoate. Riker Laboratories, Inc., Subsidiary o f 3M, St. Paul, M N, USEPA Public Docket AR-226 0458. Gilliland, F.D., 1992. Fluorocarbons and human health: studies in an occupational cohort. Ph.D. Dissertation, University o f Minnesota, Minneapolis, MN. Gilliland, F.D., Mandel, J.S., 1996. Serum perfluorooctanoic acid and hepatic enzymes, lipoproteins, and cholesterol: a study o f occupa tionally exposed men. Am. J. Ind. Med. 29, 560-568. Goecke, C.M., Jarnot, B.M., Reo, N.V., 1992. A comparative toxicological investigation o f perfluorocarboxylic acids in rats by fluorine-19 N M R spectroscopy. Chem. Res. Toxicol. 5, 512-519. Gortner, E.G., 1981. Oral teratology study o f T-2998CoC in rats. Experiment Number 0681TR0110, Safety Evaluation Laboratory and Riker Laboratories, Inc., St. Paul, M N. USEPA Public Docket AR-226-0463. Gortner, E.G., 1982. Oral teratology study o f T-3141CoC in rabbits. Experiment Number 0681TB0398, Safety Evaluation Laboratory and Riker Laboratories, Inc., St. Paul, M N. USEPA Public Docket AR-226-0465. Griffith, F.D., Long, J.E., 1980. Animal toxicity studies with ammo nium perfluorooctanoate. Am. Ind. Hyg. Assoc. J. 41, 576-583. Han, X., 2003. Ammonium perfluorooctanoate: age effect on the plasma concentration in post-weaning rats following oral gavage. Internal DuPont Project Identification DuPont-13267, dated D e cember 15, 2003. USEPA Docket AR-226. Hanhijarvi, H., Ylinen, M., Haaranen, T., Nevalainen, T., 1988. A proposed species difference in the renal excretion o f perfluorooctanoic acid in the beagle dog and rat. In: Beynen, A.C., Solleveld, H.A. (Eds.), New Developments in Biosciences: Their Implications for Laboratory Animal Science. Martinus Nijhoff Publishers, Dordrecht, Netherlands. Kalberlah, F., Schneider, K., Schuhmacher-Wolz, U., 2003. Uncer tainty in toxicological risk assessment for non-carcinogenic health hazards. Toxicol. Regul. Pharmacol. 37, 92-104. Kemper, R.A., 2003. Perfluorooctanoic acid: toxicokinetics in the rat. DuPont Haskell Laboratories, Laboratory Project ID: DuPont7473. USEPA Public Docket AR-226-1350. Kennedy Jr., G.L., 1985. Dermal toxicity o f ammonium perfluorooct anoate. Toxicol. Appl. Pharmacol. 81, 348-355. 380 J.L. Butenhoff' et al. / Regulatory Toxicology and Pharmacology 39 (2004) 363-380 Kennedy Jr., G.L., Hall, C.T., Brittelli, M.R., Barnes, J.R., Chen, H.C., 1986. Inhalation toxicity o f ammonium perfluorooctanoate. Food Chem. Toxicol. 24, 1325-1329. Kennedy Jr., G.L., Butenhoff, J.L., Olsen, G.W., O'Connor, J.C., Seacat, A.M ., Perkins, R.G., Biegel, L.B., Murphy, S.R., 2004. The toxicology o f perfluorooctanoate. Crit. Rev. Toxicol. (in press). Kerstner-Wood, C., Coward, L., Gorman, G., 2003. Protein binding of perfluorobutane sulfonate, perfluorohexanesulfonate, perfluorooctane sulfonate and perfluorooctanoate to plasma (human, rat, and monkey), and various human-derived plasma protein fractions. Southern Research Corporation, Study 9921.7. Unpublished report. USEPA Public Docket AR-226-1354. Kojo, A., Hanhijarvi, H., Ylinen, M., Kosma, V.M., 1986. Toxicity and kinetics o f perfluorooctanoic acid in the Wistar rat. Arch. Toxicol. Suppl. 9, 465-468. Kudo, N., Katakura, M., Sato, Y., Kawashima, Y., 2002. Sex hormone-regulated renal transport o f perfluorooctanoic acid. Chem. Biol. Interact. 139, 301-316. Kuslikis, B.I., Vanden Heuvel, J.P., Peterson, R.E., 1992. Lack of evidence for perfluorodecanoyl- or perfluorooctanoyl-coenzyme A formation in male and female rats. Biochem. Toxicol. 7, 25-29. Li, D., Jiao, L., 2003. Molecular epidemiology o f pancreatic cancer. Int. J. Gastrointest. Cancer 33, 3-14. Liu, R.C.M., Hahn, C., Hurtt, M.E., 1996a. The direct effect o f hepatic peroxisome proliferators on rat Leydig cell function in vitro. Fundam. Appl. Toxicol. 30, 102-108. Liu, R.C.M., Hurtt, M.E., Cook, J.C., Biegel, L.B., 1996b. Effect of the peroxisome proliferator, ammonium perfluorooctanoate (C8), on hepatic aromatase activity in adult male Crl:CD BR (CD) rats. Fundam. Appl. Pharmacol. 30, 220-228. Moore, P.S., Beghelli, S., Zamboni, G., Scarpa, A., 2003. Genetic abnormalities in pancreatic cancer. Mol. Cancer 2, 7. Mylchreest, E., 2003. PFOA: lactational and placental transport pharmacokinetic study in rats. Internal Dupont Report Identifica tion DuPont-13309, dated December 19, 2003. USEPA Public Docket AR-226. NIOSH, 2001. Health Hazard Evaluation (HETA 20010235). National Institute for Occupational Safety and Health, Medical Section, Hazard Evaluations and Technical Assistance Branch, Division of Surveillance, Hazard Evaluations and Field Studies, July 31, 2001 USEPA Public Docket AR-226-1350. Noker, P., 2003. A pharmacokinetic study o f potassium perfluorooctanoate in the cynomologus monkey. Southern Research Institute Study ID: 99214. South Research Institute, submitted to USEPA Public Docket AR-226-1362. Ohmori, K., Kudo, N., Katayama, K., Kawashima, Y., 2003. Comparison o f the toxicokinetics between perfluorocarboxylic acids with different carbon chain length. Toxicology 184, 135-140. Olsen, G.W., Gilliland, F.D., Burlew, M.M ., Burris, J.M., Mandel, J.S., Mandel, J.H., 1998. An epidemiologic investigation of reproductive hormones in men with occupational exposure to perfluorooctanoic acid. J. Occup. Environ. Med. 40, 614-620. Olsen, G.W., Burris, J.M., Burlew, M.M., Mandel, J.H., 2000. Plasma cholecystokinin and hepatic enzymes, cholesterol and lipoproteins in ammonium perfluorooctanoate production workers. Drug Chem. Toxicol. 23, 603-620. Olsen, G.W., Burris, J.M., Burlew, M.M ., Mandel, J.H., 2003a. Epidemiologic assessment o f worker serum perflurooctanesulfonate (PFOS) and perfluorooctanoate (PFOA) concentrations and medical surveillance examinations. JOEM 45, 260-270. Olsen, G.W., Hansen, K.J., Stevenson, L.A., Burris, J.M., Mandel, J.H., 2003b. Human donor liver and serum concentrations of perfluorooctanesulfonate and other perfluorochemicals. Environ. Sci. Technol. 37, 888-891. Olsen, G.W., Church, T.R., Miller, J.P., Burris, J.M., Hansen, K.J., Lundberg, J.K., Armitage, J.B., Herron, R.M ., Medhdizadehkashi, Z., Nobiletti, J.B., O'Neill, E., Mandel, J.H., Zobel, L.R., 2003c. Perfluorooctanesulfonate and other fluorochemicals in the serum o f American Red Cross adult blood donors. Environ. Health Perspect. 111, 1892-1901. Olsen, G.W., Church, T.R., Hansen, K.J., Burric, J.M., Butenhoff, J.L., Mandel, J.H., Zobel, L.R., 2004a. Quantitative evaluation of perfluorooctanesulfonate (PFOS) and other fluorochemicals in the serum o f children. J. Children's Health (in press). Olsen, G.W., Church, T.R., Larson, E.B., van Belle, G., Lundberg, J.K., Hansen, K.J., Burris, J.M., Mandel, J.H., Zobel, L.R., 2004b. Serum concentrations o f perfluorooctanesulfonate and other fluorochemicals in an elderly population from Seattle, Washington. Chemosphere 54, 1599-1611. Palazzolo, M.J., 1993. 13-Week dietary study with T-5180, ammonium perfluorooctanoate (CAS No. 3825-25-1) in male rats. Report No. HWI 6329-100, Hazleton Wisconsin, Madison, WI, USEPA Public Docket AR-226-0449. Schneider, G., Schmid, R.M., 2003. Genetic alterations in pancreatic carcinoma (Review). Mol. Cancer 2, 15. Sibinski, L.J., Allen, J.L., Erickson, E.E., 1983. Two year oral (diet) toxicity/carcinogenicity study o f fluorochemical FC-143 in rats. Experiment N o. 0281CR0012, Riker Laboratories, Inc., St. Paul, M N, USEPA Public Docket AR-226-0437, 0438, 0439, and 0440. Staples, R.E., Burgess, B.A., Kerns, W .D., 1984. The embryo-fetal toxicity and teratogenic potential o f ammonium perfluorooctanoate (APFO) in the rat. Fundam. Appl. Toxicol. 4, 429-440. Ubel, F.A., Sorenson, S.D., Roach, D.E., 1980. Health status o f plant workers exposed to fluorochemicals--a preliminary report. Am. Ind. Hyg. Assoc. J. 41, 584-589. Urrutia, R., 2002. Emerging cellular and molecular themes in pancreatic cancer research. Int. J. Gastrointest. Cancer 31, 1-4. USEPA, 1996. Proposed Guidelines for Carcinogen Risk Assessment. EPA/600/P-92/003C. Office o f Research and Development. Wash ington, DC (April, 1996). USEPA, 1999. Guidelines for Carcinogen Risk Assessment. NCEA-F0644. Risk Assessment Forum. Washington, DC. (July, 1999). USEPA, 2000. Draft Benchmark Dose Technical Guidance D ocu ment. EPA/630/R-00/001. Washington, D.C., October, 2000. USEPA, 2002. Revised draft hazard assessment o f perfluorooctanoic acid and its salts. Office o f Pollution Prevention and Toxics, Risk Assessment Division, November 4, 2002. USEPA, 2003a. Perfluorooctanoic acid (PFOA), fluorinated telomers; request for comment, solicitation o f interested parties for enforce able consent agreement development, and notice o f public meeting [OPPT-2003-0012; FRL-7303-8]. Washington, DC, April 14, 2003. USEPA, 2003b. Preliminary risk assessment o f the developmental toxicity associated with exposure to perfluorooctanoic acid and its salts. Office o f Pollution Prevention and Toxics, Risk assessment Division, April 10, 2003. USEPA, 2003c. Draft Final Guidelines for Carcinogen Risk Assess ment. EPA/630/P-03/001A, NCEA-F-0644A, Washington, DC, February, 2003. USEPA, 2004. Request for Nominations for Experts for Perfluorooctanoic Acid Human Health Assessment Review Panel. Federal Register 69, 16249-16250 (March 29, 2004). Vanden Heuvel, J., Kuslikis, B., Van Rafelghem, M., 1991. Tissue distribution, metabolism, and elimination o f perfluorooctanoic acid in male and female rats. Biochem. Toxicol. 6, 83-92. Wagner, J.W., 1975. Fundamentals o f Clinical Pharmacokinetics. Drug Intelligence Press, Hamilton, IL. p. 129. West Virginia Department o f Environmental Protection, 2002. Final ammonium perfluorooctanoate (C8) assessment o f toxicity team (CATT) report. August, 2002. Ylinen, M., Kojo, A., Hanhijarvi, H., Peura, P., 1990. Deposition of perfluorooctanoic acid in the rat after single and subchronic administration. Bull. Environ. Contam. Toxicol. 44, 46-53.